NASA: Radar Ice Altimetry

Investigation of the Observed ERS-1/ERS-2 Bias

Investigation of the ERS-2 and ERS-1 IDRs in the ocean near Greenland inferred that there was a significant bias of around 40cm between elevations calculated from ranges from the two satellites with ERS-2 being lower. This is very close to the documented difference in the OPR data for ERS-1 and ERS-2 of 40.9 cm due to the differences in the eps-tau-g correction (Femenias 1997, "ERS QLOPR and OPR Range Processing",ER-TN-RS-RA -0022) though we have found no documentation stating that this bias exists for ERS-1 and ERS-2 data processed from the WAP.

To validate that this bias was not regional, we processed global ocean data from the WAP ERS-1 for January and July from 1993-1996 for ERS-1 and for 1995-1996 for ERS-2. All the like-mission crossover statistics show means of 0-2+/-20 cm whereas the cross-mission statistics show means of 30+/-17 cm with ERS2 being lower. The data processing was exactly the same as for over the ice sheets except that we applied an EM bias of 5.5% of the SWH to both missions. This showed ERS2 WAP ocean ranges had a significant bias with respect to ERS1 WAP ranges, however it was not the expected 40-41cm. This may be due to the fact that we did not use optimal ocean processing. Optimal ocean processing would include using the radiometer correction instead of a model for the wet troposphere delay, applying a barotropic correction, using a more accurate em bias calculation, and better editing.

We also compared one pass each of ERS-1 WAP and ERS-2 WAP data ove r the Arctic Ocean to the respective OPR data. We compared the measurement values after internal and Doppler corrections were made. The OPR measurement also had the OCOG retracking correction applied, but the OPR does not give the actual value used. An article written in 1994 by Johnathan Bamber stated that they used the 50% OCOG and a look up table. The look up table value is given on the OPR and is set to zero for these two passes.

The mean of the difference between the ERS-1 WAP measurement and the OPR was 6.8+/- 17.5 cm when the OCOG 50% retracker was used and 4.1+/-25.3 cm when the GSFC V4 retracker was used on the WAP measurement. For ERS-2 the mean difference was 7.9+/-10.5cm for the OCOG 50% retracker and 1.7+/-14.3 cm for the GSFC V4 retracker. If the ERS-1 and ERS-2 WAP data were processed the same way we would have expected close to a 40 cm mean between the ERS-2 WAP and the OPR measurement. Our results infer that the ERS-2 WAP has some correction different than the ERS-1 WAP. We suspect that the eps-tau-g correction was modified for ERS-2 WAP based on the better ERS-2 ground calculation.

The eps-tau-g recharacterization was confirmed in an E-mail message from Pierre Femenias. The text of this message follows:

Dear Anita,
Please find hereafter information on the ERS Radar Altimeter external range biases which I hope will answer to your query from July 27 2000 regarding the WAP data.

As developed in the note mentioned here above, the determination of the ERS-2 characterisation data has been performed with more reliable techniques leading to different parameters as it is the case for the eps_tau_g parameter. The difference between ERS-1 and ERS-2 is very close to the Venice bias and it is for this reason that most of the users apply this value (40.9 cm) as external bias to the ERS-1 data instead of the Venice one.